ARDS Subphenotypes: Differential Treatment Responses, and

Where Do We Go From Here?

Carolyn S. Calfee, MD MAS

Associate Professor of Medicine and Anesthesia

University of California, San Francisco

Critical Care Canada Forum

November 8, 2018

Research funding: NIH, US Food and Drug Administration, GlaxoSmithKline, Bayer

Medical advisory board/consulting: GlaxoSmithKline, BoehringerIngelheim, Bayer, Roche/Genentech, Prometic, CSL Behring

Disclosures

Are Pharmacotherapy Trials in ARDS Doomed?

Corticosteroids

Surfactant

Prostaglandin E1

Anti-endotoxins

Anti-cytokines

Procysteine

Ketoconazole

Lisofylline

Activated protein C

Beta-agonists

Omega-3 fatty acids

* A partial list

Decades of negative pharmacotherapy trials for ARDS,* despite promising mechanisms and preclinical data

Neutrophil elastase inhibitor

Statins

Nitric oxide

Ibuprofen

ARDS

“Enrichment”: Focus on patient population in which detection of drug effect is more likely than in unselected population

Prognostic Enrichment: Focus on patients at highest risk for disease-related endpoint (i.e. death due to ARDS)

Increase absolute effect size, not relative effect

Predictive Enrichment: Focus on patients most likely to respond to a particular treatment

Enrich for a specific endotype relevant to mechanism

Enrichment: Two Strategies For Improving Clinical Trials

FDA Draft Guidance to Industry, December 2012; Prescott AJRCCM 2016

ARDS is a syndrome defined by non-specific criteria

Clinical, biological, pathologic heterogeneity

If we split rather than lump, can we find a treatment-responsive group?

How To Move Towards Enrichment in ARDS?

Prognostic Enrichment in ARDS: Some Successes

Wenzel S, Nature Med 2012

Asthma: Not Just a Single Disease

One Approach To Finding Subtypes: Latent Class Analysis (LCA)

Based on hypothesis that there are distinct sub-classes within a broader group of patients

Does data distribution better fit with 1 class or 2 (3, 4, …)?

Tests hypothesis re: fit of number of classes

Latent Class Analysis Finds Two Subphenotypes of ARDS in RCT’s

Famous et al, AJRCCM 2017

Comparison of Two ARDS Subphenotypes

“Hyper-inflammatory”

“Hypo-inflammatory”

Subphenotype Membership is Stable Over Several Days

In both ARMA (low TV) and ALVEOLI (low vs high PEEP), two classes present on Study Day 3

In both studies, 94% patients stayed in same class from Day 0 to 3

In ALVEOLI, Probability of Class Transition (sample size)

Day 3

Class 1 Class 2

Day 0 Class 1 0.95 (313) 0.05 (18)

Class 2 0.11 (11) 0.89 (138)

Delucchi et al, Thorax 2018

ARDS Subphenotypes Are Similar At Days 0 And 3

Delucchi et al, Thorax 2018

90d Mortality Much Worse in Hyper-inflammatory Subphenotype

TrialHypo-

inflammatoryHyper-

inflammatoryP-value

ARMA (low TV) 23% 44% 0.006

ALVEOLI (PEEP) 19% 51% <0.001

FACTT (fluids) 22% 45% <0.0001

HARP2 (simvastatin) 22% 47% <0.0001

SAILS (rosuvastatin) 21% 38% <0.0001

In secondary analyses of RCT’s, two subphenotypesresponded differentially to:

Mechanical ventilation: High vs low PEEP

Fluids: Conservative vs. liberal

Pharmacotherapy: Simvastatin

NOT rosuvastatin

ARDS Subphenotypes: Differential Response to Therapies

Original trial of Simvastatin: HARP-2

McAuley NEJM 2014

Hyperinflammatory Subphenotype: Survival Benefit from Simvastatin

28 Day Survival

* Secondary analysis: Needs prospective validationCalfee Lancet Resp 2018

No Evidence of Benefit From Rosuvastatin

Sinha et al, ICM 2018

Dose-related?

Lipophilic vs. hydrophilic?

Simvastatin vs. Rosuvastatin

Singla et al, Pulm Circ 2012

Diffuse vs. Focal ARDS: Might Different Ventilatory Strategies Work?

Mrozek et al, CHEST 2016; clinicaltrials.gov

• Lung Imaging for Ventilatory Setting in ARDS (LIVE) trial

• 420 patients• Randomized to control (low TV

ventilation) vs. MV tailored to radiographic pattern:• Focal: Early proning, low PEEP• Diffuse: Recruitment

maneuver, high PEEP • Completed enrollment

• No results available

Aspirations for ARDS: Targeting Specific Subgroups

Chapman et al, NEJM 2011

Fisher C et al, JAMA 1994; Meyer N, CCM 2018

A Cautionary Tale: Recombinant IL1RA in Sepsis

Plasma IL1RA low

Pro

po

rtio

n A

live

Plasma IL1RA highrhIL1RA

placebo

rhIL1RA

placebo

Days

Meyer N, CCM 2018

p=0.044 for interaction

A Cautionary Tale: Recombinant IL1RA in Sepsis

Thompson BT et al, NEJM 2017

Targeted Clinical Trials: How To Get There From Here?

Identification of subgroups; retrospective analysis of differential treatment response in RCTs

Targeted clinical trials

Identification of subgroups; retrospective analysis of differential treatment response in RCTs

Prospective identification

Understanding of mechanistic drivers

More populations, more subgroups?

Targeted clinical trials

Genes? Environment? Host vs. insult

Targeted Clinical Trials: How To Get There From Here?

Adaptive clinical trials?

Prognostic and/or predictive enrichment may help find the responsive subset

First steps: Understanding that there is heterogeneity in ARDS and it may affect treatment response Retrospective identification of treatment-responsive subgroups

Survival benefit in hyper-inflammatory subphenotype with simvastatin but not rosuvastatin

Next steps: Prospective studies are required

Understanding mechanisms will be fundamental

Conclusions